1 APACHE 2.5 STATUS: -*-text-*-
2 Last modified at [$Date$]
4 The current version of this file can be found at:
6 * http://svn.apache.org/repos/asf/httpd/httpd/trunk/STATUS
8 Documentation status is maintained separately and can be found at:
10 * docs/STATUS in this source tree, or
11 * http://svn.apache.org/repos/asf/httpd/httpd/trunk/docs/STATUS
13 Consult the following STATUS files for information on related projects:
15 * http://svn.apache.org/repos/asf/apr/apr/trunk/STATUS
17 Patches considered for backport are noted in their branches' STATUS:
19 * http://svn.apache.org/repos/asf/httpd/httpd/branches/2.0.x/STATUS
20 * http://svn.apache.org/repos/asf/httpd/httpd/branches/2.2.x/STATUS
21 * http://svn.apache.org/repos/asf/httpd/httpd/branches/2.4.x/STATUS
26 [NOTE that x.{odd}.z versions are strictly Alpha/Beta releases,
27 while x.{even}.z versions are Stable/GA releases.]
29 -8s 2.5.0 : In Development.
31 Contributors looking for a mission:
33 * Just do an egrep on "TODO" or "XXX" in the source.
35 * Review the bug database at: http://issues.apache.org/bugzilla/
37 * Review the "PatchAvailable" bugs in the bug database:
39 https://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
41 After testing, you can append a comment saying "Reviewed and tested".
43 * Open bugs in the bug database.
45 * See also the STATUS file in the docs/ directory, which lists documentation-specific TODO items.
48 CURRENT RELEASE NOTES:
57 THINGS THAT SHOULD BE CONSIDERED EARLY IN THE 2.6/3.0 DEVELOPMENT CYCLE:
59 * Seriously ramp up/replace test framework and cases to have better
60 coverage of existing special cases and behaviours users rely on.
62 * Add performance testing to the test framework.
64 * Competely untangle core filesystem behavior where a filesystem htdocs/
65 resource wasn't indicated by the request URI.
67 * Refactor r->uri into a %escaped raw form presented by the client, and
68 a distinct decoded field used only for local filesystem access.
70 * Change default prefix from /usr/local/apache2 to something corresponding
71 to the project name. Rename apachectl to httpdctl.
73 * Change merge order of <Location> to be most specific match last. This
74 is more consistent with <Directory> and allows some optimizations for the
77 * Detect Lua 5.2.0 during configure and add LUA_COMPAT_ALL to CPPFLAGS.
78 Maybe it even suffices to add LUA_COMPAT_MODULE and individually
79 care about the two remaining incompatible code lines (one with lua_strlen,
82 * Event's timeout_mutex to enter keepalive state probably needs some
85 * Better H2 integration?
86 - adding handling of slave connections to mpm, no extra H2 workers,
87 triggering "events" read/write/timer from main/slave
88 - add slave writes/done/abort to events that wake up master connection
89 - disentangle core filters to server one purpose only, so that H2
90 versions can reuse them properly.
92 * Remove mod_access_compat?
94 * Ditch platforms/89/old prereqs or anything else?
96 * Leverage libmill? Drop serf?
98 * Better abstraction of slave connections and "requests".
99 - add abstraction for "response" as something that can be passed
100 through filters. To be serialized into the correct HTTP bytes on
102 - solve multi-threaded access to master connection props/module conf
105 * make mod_ssl more "core"?
107 * add high-level server configuration directives that can steer/influence
108 module defaults/warn/rejects related to security
110 * Ditch HTTP/0.9? At least, make HttpProtocolOptions Require1.0 the default.
112 * Restructure merge fn table/indexes to ignore modules with no directives,
113 and permit modules with dozens upon dozens of merge values to split these
114 into multiple functional config groups to avoid excessive merging.
115 Retitle from 'per-dir' to 'per-location' to better reflect the always-run
116 sections (location, ifexpr etc), while we phase out the file-oriented
119 * New versioning or release cadence.
121 * Ditch old APIs when we have the chance with 3.x. Consolidate current
122 functionality into APIs with stronger guarantees. (Specific examples TBD.)
124 * Remove as many undesirable-but-kept-for-backwards-compatibility behaviors
125 as possible from current config directives. (Specific examples TBD.)
127 * True event-loop/asynchronous support in the server core.
129 * Modify configuration syntax to separate meta-directives from runtime
130 directives (e.g. If vs. IfVersion). Allow as much static analysis of the
131 configuration as possible without needing to start the server to figure out
134 * Support JSON-like configuration files
136 * Opaque data structures w/ getters/setters
138 * Generic interface to enable runtime changes (adjusting log level, modifying
139 balancer information, toggling flags on/off). Perhaps modules can register
140 callbacks for making these changes?
142 * REST-based administration for existing (balancer/etc) and new dynamic
143 runtime changes (see above)
145 * Improve the look of generated pages (status, load-balancer...) with dynamic
146 update of the values. Generate HTML5 pages, instead of 3.2, Get rid of XHTML
147 in the generated pages.
149 * Add performance monitoring of the server, of each module (?), in order to help
150 understanding what worth looking at in order to improve overall performance.
151 (https://cdn.wp.nginx.com/wp-content/uploads/2016/12/Amplify-Dashboards-page-base-for-filters.png)
153 * Drop CGI-1.1-incompatible behaviors kept for compatibility reasons with
154 "broken" server implementations (PR 51517). (Note that many of them are
155 "broken" *because* of our behaviors.)
158 OLD ISSUES THAT WERE THOUGHT TO BE SHOWSTOPPERS FOR 2.4 BUT OBVIOUSLY WEREN'T:
160 * Handling of non-trailing / config by non-default handler is broken
161 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105451701628081&w=2
162 jerenkrantz asks: Why should this block a release?
163 wsanchez agrees: this may be a change in behavior, but isn't
164 clearly wrong, and even if so, it doesn't seem like a
167 * the edge connection filter cannot be removed
168 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105366252619530&w=2
169 http://mail-archives.apache.org/mod_mbox/httpd-dev/200501.mbox/%3C41E30B42.4060202@stason.org%3E
170 jerenkrantz asks: Why should this block a release?
171 stas replies: because it requires a rewrite of the filters stack
172 implementation (you have suggested that) and once 2.2 is
173 released you can't do that anymore.
174 pgollucci: this affects mod_perl I'm pretty sure.
177 RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
179 * Clean up all the kruft and *extremely* outdated stuff below...
181 * Maybe remove Limit/LimitExcept or at least make it log warnings when
184 * Patches submitted to the bug database:
185 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
187 * Filter stacks and subrequests, redirects and fast redirects.
188 There's at least one PR that suffers from the current unclean behaviour
189 (which lets the server send garbage): PR 17629
190 nd says: Every subrequest should get its own filter stack with the
191 subreq_core filter as bottom-most. That filter does two things:
192 - swallow EOS buckets
193 - redirect the data stream to the upper request's (rr->main)
194 filter chain directly after the subrequest's starting
196 Once we have a clean solution, we can try to optimize
197 it, so that the server won't be slow down too much.
199 * RFC 2616 violations.
200 Closed PRs: 15852, 15857, 15859, 15861, 15864, 15869, 15870, 16120,
201 16125, 16135, 16136, 16137, 16138, 16139, 16140, 16518,
203 Open PRs: 15865, 15866, 15868, 16126, 16133, 16142, 16521, 42978
204 jerenkrantz says: need to decide how many we need to backport and/or
205 if these rise to showstopper status.
206 wrowe suggests: it would be nice to see "MUST" v.s. "SHOULD" v.s. "MAY"
207 out of this list, without reviewing them individually.
208 wrowe asks: what is lingering after 2.4.25 release? Offhand, only
211 * pipes deadlock on all platforms with limited pipe buffers (e.g. both
212 Linux and Win32, as opposed to only Win32 on 1.3). The right solution
213 is either GStein's proposal for a "CGI Brigade", or OtherBill's proposal
214 for "Poll Buckets" for "Polling Filter Chains". Or maybe both :-)
216 * All handlers should always send content down even if r->header_only
217 is set. If not, it means that the HEAD requests don't generate the
218 same headers as a GET which is wrong.
220 * exec cmd and suexec arg-passing enhancements
221 Status: Patches proposed
222 Message-ID: <20020526041748.A29148@prodigy.Redbrick.DCU.IE>
223 (see the "proc.patch" and "suexec-shell.patch" links in this message)
225 * The 2.0.36 worker MPM graceless shutdown changes work but are
226 a bit clunky on some platforms; eg, on Linux, the loop to
227 join each worker thread seems to hang, and the parent ends up
228 killing off the child with SIGKILL. But at least it shuts down.
230 chrisd: Has this been fixed by the changes for PR 38737?
232 * We do not properly substitute the prefix-variables in the configuration
233 scripts or generated-configs. (i.e. if sysconfdir is etc,
234 httpd-std.conf points to conf.)
236 * If any request gets through ap_process_request_internal() and is
237 scheduled to be served by the core handler, without a flag that this
238 r->filename was tested by dir/file_walk, we need to 500 at the very
239 end of the ap_process_request_internal() processing so sub_req-esters
240 know this request cannot be run. This provides authors of older
241 modules better compatibility, while still improving the security and
244 Status: still need to decide where this goes, OtherBill comments...
245 Message-ID: <065701c14526$495203b0$96c0b0d0@roweclan.net>
246 [Deleted comments regarding the ap_run_handler phase, as irrelevant
247 as BillS points out that "common case will be caught in
248 default_handler already (with the r->finfo.filetype == 0 check)"
249 and the issue is detecting this -before- we try to run the req.]
251 gregames says: can this happen somehow without a broken module
252 being involved? If not, why waste cycles trying to defend against
253 potential broken modules? It seems futile.
254 wrowe counters: no, it shouldn't happen unless the module is broken.
255 But the right answer is to fail the request up-front in dir/file
256 walk if the path was entirely invalid; and we can't do that either
257 UNTIL 2.1 or we break modules that haven't hooked map_to_storage.
259 * Can a static httpd be built reliably?
260 Message-ID: <20020207142751.T31582@clove.org>
262 * Usage of APR_BRIGADE_NORMALIZE in core_input_filter should be
265 <Pine.LNX.4.33.0201202232430.318-100000@deepthought.cs.virginia.edu>
266 Jeff wonders if we still care about this. It is no longer an
267 API issue but simply an extra trip through the brigade.
269 * Try to get libtool inter-library dependency code working on AIX.
270 Message-ID: <cm3n10lx555.fsf@rdu163-40-092.nc.rr.com>
272 Justin says: If we get it working on AIX, we can enable this
273 on all platforms and clean up our build system somewhat.
274 Jeff says: I thought I tested a patch for you sometime in
275 January that you were going to commit within a few days.
277 * Handling of %2f in URIs. Currently both 1.3 and 2.0
278 completely disallow %2f in the request URI path (see
279 ap_unescape_url() in util.c). It's permitted and passed
280 through in the query string, however. Roy says the
281 original reason for disallowing it, from five years ago,
282 was to protect CGI scripts that applied PATH_INFO to
283 a filesystem location and which might be tricked by
284 ..%2f..%2f(...). We *should* allow path-info of the
285 form 'http://foo.com/index.cgi/path/to/path%2finfo'.
286 Since we've revamped a lot of our processing of path
287 segments, it would be nice to allow this, or at least
288 allow it conditionally with a directive.
290 OtherBill adds that %2f as the SECOND character of a multibyte
291 sequence causes the request to fail! This happens notably in
294 * There is increasing demand from module writers for an API
295 that will allow them to control the server à la apachectl.
296 Reasons include sole-function servers that need to die if
297 an external dependency (e.g., a database) fails, et cetera.
298 Perhaps something in the (ever more abused) scoreboard?
300 On the other hand, we already have a pipe that goes between parent
301 and child for graceful shutdown events, along with an API that
302 can be used to send a message down that pipe. In threaded MPMs,
303 it is easy enough to make that one pipe be used for graceful
304 and graceless events, and it is also easy to open that pipe
305 to both parent and child for writing. Then we just need to
306 figure out how to do graceless on non-threaded MPMs.
308 * Allow the DocumentRoot directive within <Location > scopes? This
309 allows the beloved (crusty) Alias /foo/ /somepath/foo/ followed
310 by a <Directory /somepath/foo> to become simply
311 <Location /foo/> DocumentRoot /somefile/foo (IMHO a bit more legible
312 and in-your-face.) DocumentRoot unset would be accepted [and would
313 not permit content to be served, only virtual resources such as
314 server-info or server-status.
315 This proposed change would _not_ depricate Alias.
316 striker: See the thread starting with Message-ID:
317 JLEGKKNELMHCJPNMOKHOGEEJFBAA.striker@apache.org.
319 * Win32: Rotatelogs sometimes is not terminated when Apache
320 goes down hard. FirstBill was looking at possibly tracking the
321 child's-child processes in the parent process.
322 stoddard: Shared scoreboard might offer a good way for the parent
323 to keep track of 'other child' processes and whack them if the child
325 Other thoughts on walking the process chain using the NT kernel
326 have also been proposed on APR.
328 * Eliminate unnecessary creation of pipes in mod_cgid
330 * Combine log_child and piped_log_spawn. Clean up http_log.c.
333 * Platforms that do not support fork (primarily Win32 and AS/400)
334 Architect start-up code that avoids initializing all the modules
335 in the parent process on platforms that do not support fork.
337 * There are still a number of places in the code where we are
338 losing error status (i.e. throwing away the error returned by a
339 system call and replacing it with a generic error code)
341 * Mass vhosting version of suEXEC.
343 * All DBMs suffer from confusion in support/dbmmanage (perl script) since
344 the dbmmanage employs the first-matched dbm format. This is not
345 necessarily the library that Apache was built with. Aught to
346 rewrite dbmmanage upon installation to bin/ with the proper library
347 for predictable mod_auth_dbm administration.
348 Questions; htdbm exists, time to kill dbmmanage, or does it remain
349 useful as a perl dbm management example? If we keep it,
350 do we address the issue above?
353 Some additional items remaining:
354 - case_preserved_filename stuff
355 (use the new canonical name stuff?)
356 - find a new home for ap_text(_header)
357 - is it possible to remove the DAV: namespace stuff from util_xml?
359 * ap_core_translate() and its use by mod_mmap_static and mod_file_cache
360 are a bit wonky. The function should probably be exposed as a utility
361 function (such as ap_translate_url2fs() or ap_validate_fs_url() or
362 something). Another approach would be a new hook phase after
363 "translate" which would allow the module to munge what the
364 translation has decided to do.
365 Status: Greg +1 (volunteers)
367 * Explore use of a post-config hook for the code in http_main.c which
368 calls ap_fixup_virutal_hosts(), ap_fini_vhost_config(), and
369 ap_sort_hooks() [to reduce the logic in main()]
371 * read the config tree just once, and process N times (as necessary)
373 * (possibly) use UUIDs in mod_unique_id and/or mod_usertrack
375 * (possibly) port the bug fix for PR 6942 (segv when LoadModule is put
376 into a VirtualHost container) to 2.0.
378 * shift stuff to mod_core.h
380 * callers of ap_run_create_request() should check the return value
381 for failure (Doug volunteers)
383 * Fix the worker MPM to use POD to kill child processes instead
384 of ap_os_killpg, regardless of how they should die.
386 chrisd: Is this done, by any chance? See r92598 and r93358.
388 * Scoreboard structures could be changed in the future such that
389 proper alignment is not maintained, leading to segfaults on
390 some systems. Cliff posted a patch to deal with this issue but
391 later recanted. See this message to dev@apr.apache.org:
393 <Pine.LNX.4.44.0203011354090.16457-200000@deepthought.cs.virginia.edu>
395 * APXS either needs to be fixed completely for use when apr is out of tree,
396 or it should drop query mode altogether, and we just grow an
397 httpd-config or similar arrangement.
398 To quote a discussion in STATUS earlier:
400 thommay: this doesn't fix all the problems with apxs and out of
401 tree apr/apr-util, but it's a good start. There's still the
402 query cases; but I'm beginning to think that in these cases
403 the app should be querying ap{r,u}-config directly
404 deprecate -q: add htpd-config: gstein, pquerna, minfrin, pgollucci
407 TODO ISSUES REMAINING IN MOD_SSL:
409 * SSLRequire directive (parsing of) leaks memory
411 * ssl_var_lookup could be rewritten to be MUCH faster
414 * mod_proxy: Ability to run SSL over proxy gateway connections,
415 encrypting (or reencrypting) at the proxy.
417 * mod_cache: Handle ESI tags.
419 * mod_cache: Resolve issue of how to cache page fragments (or perhaps
420 -if- we want to cache page fragments). Today, mod_cache/mod_mem_cache
421 will cache #include 'virtual' requests (but not #include 'file'
422 requests). This was accomplished by making CACHE_IN a
423 CONTENT_SET-1 filter to force it to run before the SUBREQ_CORE
424 filter. But now responses cannot be cached that include the
425 effects of having been run through CONTENT_SET filters
426 (mod_deflate, mod_expires, etc). We could rerun all the
427 CONTENT_SET filters on the cached response, but this will not
428 work in all cases. For example, mod_expires relies on installing
429 the EXPIRATION filter during fixups. Contents served out of
430 mod_cache (out of the quick_handler) bypass -all- the request
431 line server hooks (Ryan really hated this. It is great for
432 performance, but bad because of the complications listed above).
434 mod_cache/mod_mem_cache/mod_cache_disk:
436 * mod_mem_cache: Consider adding a RevalidateTimeout directive to
437 specify time at which local cached content is to be revalidated
438 (ie, underlying file stat'ed to see if it has changed).
440 * mod_mem_cache/mod_cache_disk: Need to be able to query cache
441 status (num of entries, cache object properties, etc.).
442 mod_status could be extended to query optional hooks defined
443 by modules for the purpose of reporting module status.
444 mod_cache (et. al.) could define optional hooks that are called
445 to collect status. Status should be queryable by
447 jerenkrantz says: Yawn. Who cares.
449 * Regex containers don't work in an intutive way
450 Status: No one has come up with an efficient way to fix this
451 behavior. Dean has suggested getting rid of regex containers
453 OtherBill suggests: We at least seem to agree on eliminating
454 the <Container ~ foo> forms, and using only
455 <ContainerMatch foo> semantics.
457 * orig_ct in the byterange/multipart handling may not be
458 needed. Apache 1.3 just never stashed "multipart" into
459 r->content_type. We should probably follow suit since the
460 byterange stuff doesn't want the rest of the code to see the
461 multipart content-type; the other code should still think it is
462 dealing with the <orig_ct> stuff.
463 Status: Greg volunteers to investigate (esp. since he was most
464 likely the one to break it :-)
466 EXPERIMENTAL MODULES:
468 Experimental modules should eventually be be promoted to fully supported
469 status or removed from the repository entirely (ie, the
470 'experiment' failed). This section tracks what needs to happen to
471 get the modules promoted to fully supported status.